EBIC

EBIC performs evolutionary-based biclustering to identify subsets of genes and experimental conditions with coherent patterns in high-dimensional genomic and epigenetic datasets.

Key Features:

• GPU Acceleration: Full support for multiple GPUs with parallel execution and demonstrated scalability, including over 6.6-fold speedup on a cluster of eight GPUs versus a single GPU.
• R and Bioconductor Integration: Interoperability with the R programming environment and Bioconductor for incorporation into R-based analysis workflows.
• Handling Missing Values: Option to exclude missing values from analyses to manage incomplete datasets.

Scientific Applications:

• Genomic data mining: Discovery of coherent gene–condition biclusters in large-scale genomic datasets.
• Epigenetic analysis: Analysis of DNA methylation datasets, including demonstrated use on a dataset with 436,444 rows.
• High-dimensional expression profiling: Extraction of complex expression patterns across genes and samples that may be missed by traditional clustering.

Methodology:

An evolutionary-based biclustering algorithm identifies subsets of genes and conditions with similar expression profiles, includes an option to exclude missing values, and supports multi-GPU parallel execution.